How the number is calculated
For each category, multiply tokens by its USD rate per million, then divide by 1,000,000. Add the four category amounts. For example, 1 million uncached input tokens at $2 per million and 100,000 output tokens at $10 per million give $3.00 before any other token categories or charges. These are illustrative rates, not current prices for a named model.
Where to find a rate
Use your provider's current price for the exact model, processing tier, context size, region, and token category. OpenAI's API pricing and Claude API pricing distinguish input, cached input or writes, and output. Some plans or tools have other charges and discounts. Check the provider's bill for the amount actually charged.
What this cannot tell you
A subscription limit is not an API invoice. Canopy's usage view can summarize supported CLI data, but different CLIs expose different categories and timestamps. This worksheet does not include tools, long-context or regional multipliers, taxes, bundled usage, rebates, or unreported work. A zero result with missing categories means insufficient data, not a free session.
To optimize a task, compare accepted code, retries, review time, and the final result with the estimate. Read the token-usage guide and why a cost dashboard is not a bill.
Frequently asked questions
Does Canopy charge this amount?
No. The calculator does not contact Canopy or a model provider. It computes a hypothetical API-equivalent amount from what you enter.
Can I paste the 'sent' and 'received' totals from any CLI?
Only if you know how that CLI maps them to the provider's billed token categories. Cached input, cache writes, and subscription plans can change the interpretation.